bos@149: #!/usr/bin/env python
bos@251: #
bos@251: # This script attempts to work around some of the more bizarre and
bos@251: # quirky behaviours of htlatex.
bos@251: #
bos@251: # - We've persuaded htlatex to produce UTF-8, which unfortunately
bos@251: #   causes it to use huge character sequences to represent even the
bos@251: #   safe 7-bit ASCII subset of UTF-8.  We fix that up.
bos@251: #
bos@251: # - BUT we have to treat angle brackets (for example, redirections in
bos@251: #   shell script snippets) specially, otherwise they'll break the
bos@251: #   generated HTML.  (Reported by Johannes Hoff.)
bos@251: #
bos@251: # - For some reason, htlatex gives a unique ID to each fancyvrb
bos@251: #   environment, which makes writing a sane, small CSS stylesheet
bos@251: #   impossible.  We squish all those IDs down to nothing.
bos@149: 
bos@149: import os
bos@149: import sys
bos@149: import re
bos@149: 
bos@251: angle_re = re.compile(r'(&#x003[CE];)')
bos@251: unicode_re = re.compile(r'&#x00([0-7][0-9A-F]);')
bos@149: fancyvrb_re = re.compile(r'id="fancyvrb\d+"', re.I)
bos@260: ligature_re = re.compile(r'&#xFB0([0-4]);')
bos@149: 
bos@149: tmpsuffix = '.tmp.' + str(os.getpid())
bos@149: 
bos@251: def hide_angle(m):
bos@251:     return m.group(1).lower()
bos@251: 
bos@149: def fix_ascii(m):
bos@149:     return chr(int(m.group(1), 16))
bos@149: 
bos@260: ligatures = ['ff', 'fi', 'fl', 'ffi', 'ffl']
bos@260: 
bos@260: def expand_ligature(m):
bos@260:     return ligatures[int(m.group(1))]
bos@260: 
bos@149: for name in sys.argv[1:]:
bos@149:     tmpname = name + tmpsuffix
bos@149:     ofp = file(tmpname, 'w')
bos@149:     for line in file(name):
bos@251:         line = angle_re.sub(hide_angle, line)
bos@149:         line = unicode_re.sub(fix_ascii, line)
bos@260:         line = ligature_re.sub(expand_ligature, line)
bos@149:         line = fancyvrb_re.sub('id="fancyvrb"', line)
bos@149:         ofp.write(line)
bos@149:     ofp.close()
bos@149:     os.rename(tmpname, name)